=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.350 retrieving revision 1.353 diff -u -p -r1.350 -r1.353 --- mandoc/roff.c 2018/12/14 05:18:03 1.350 +++ mandoc/roff.c 2018/12/18 22:00:02 1.353 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.350 2018/12/14 05:18:03 schwarze Exp $ */ +/* $Id: roff.c,v 1.353 2018/12/18 22:00:02 schwarze Exp $ */ /* * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons * Copyright (c) 2010-2015, 2017, 2018 Ingo Schwarze @@ -98,7 +98,6 @@ struct mctx { }; struct roff { - struct mparse *parse; /* parse point */ struct roff_man *man; /* mdoc or man parser */ struct roffnode *last; /* leaf of stack */ struct mctx *mstack; /* stack of macro contexts */ @@ -780,12 +779,11 @@ roff_free(struct roff *r) } struct roff * -roff_alloc(struct mparse *parse, int options) +roff_alloc(int options) { struct roff *r; r = mandoc_calloc(1, sizeof(struct roff)); - r->parse = parse; r->reqtab = roffhash_alloc(0, ROFF_RENAMED); r->options = options; r->format = options & (MPARSE_MDOC | MPARSE_MAN); @@ -844,13 +842,11 @@ roff_man_free(struct roff_man *man) } struct roff_man * -roff_man_alloc(struct roff *roff, struct mparse *parse, - const char *os_s, int quick) +roff_man_alloc(struct roff *roff, const char *os_s, int quick) { struct roff_man *man; man = mandoc_calloc(1, sizeof(*man)); - man->parse = parse; man->roff = roff; man->os_s = os_s; man->quick = quick; @@ -1158,6 +1154,7 @@ roff_res(struct roff *r, struct buf *buf, int ln, int struct roff_node *n; /* used for header comments */ const char *start; /* start of the string to process */ char *stesc; /* start of an escape sequence ('\\') */ + const char *esct; /* type of esccape sequence */ char *ep; /* end of comment string */ const char *stnam; /* start of the name, after "[(*" */ const char *cp; /* end of the name, e.g. before ']' */ @@ -1167,7 +1164,6 @@ roff_res(struct roff *r, struct buf *buf, int ln, int size_t naml; /* actual length of the escape name */ size_t asz; /* length of the replacement */ size_t rsz; /* length of the rest of the string */ - enum mandoc_esc esc; /* type of the escape sequence */ int inaml; /* length returned from mandoc_escape() */ int expand_count; /* to avoid infinite loops */ int npos; /* position in numeric expression */ @@ -1176,6 +1172,7 @@ roff_res(struct roff *r, struct buf *buf, int ln, int int done; /* no more input available */ int deftype; /* type of definition to paste */ int rcsid; /* kind of RCS id seen */ + enum mandocerr err; /* for escape sequence problems */ char sign; /* increment number register */ char term; /* character terminating the escape */ @@ -1308,7 +1305,10 @@ roff_res(struct roff *r, struct buf *buf, int ln, int term = '\0'; cp = stesc + 1; - switch (*cp) { + if (*cp == 'E') + cp++; + esct = cp; + switch (*esct) { case '*': case '$': res = NULL; @@ -1324,12 +1324,26 @@ roff_res(struct roff *r, struct buf *buf, int ln, int res = ubuf; break; default: - esc = mandoc_escape(&cp, &stnam, &inaml); - if (esc == ESCAPE_ERROR || - (esc == ESCAPE_SPECIAL && - mchars_spec2cp(stnam, inaml) < 0)) - mandoc_msg(MANDOCERR_ESC_BAD, - ln, (int)(stesc - buf->buf), + err = MANDOCERR_OK; + switch(mandoc_escape(&cp, &stnam, &inaml)) { + case ESCAPE_SPECIAL: + if (mchars_spec2cp(stnam, inaml) >= 0) + break; + /* FALLTHROUGH */ + case ESCAPE_ERROR: + err = MANDOCERR_ESC_BAD; + break; + case ESCAPE_UNDEF: + err = MANDOCERR_ESC_UNDEF; + break; + case ESCAPE_UNSUPP: + err = MANDOCERR_ESC_UNSUPP; + break; + default: + break; + } + if (err != MANDOCERR_OK) + mandoc_msg(err, ln, (int)(stesc - buf->buf), "%.*s", (int)(cp - stesc), stesc); stesc--; continue; @@ -1386,7 +1400,7 @@ roff_res(struct roff *r, struct buf *buf, int ln, int cp++; break; } - if (*cp++ != '\\' || stesc[1] != 'w') { + if (*cp++ != '\\' || *esct != 'w') { naml++; continue; } @@ -1394,6 +1408,7 @@ roff_res(struct roff *r, struct buf *buf, int ln, int case ESCAPE_SPECIAL: case ESCAPE_UNICODE: case ESCAPE_NUMBERED: + case ESCAPE_UNDEF: case ESCAPE_OVERSTRIKE: naml++; break; @@ -1407,7 +1422,7 @@ roff_res(struct roff *r, struct buf *buf, int ln, int * undefined, resume searching for escapes. */ - switch (stesc[1]) { + switch (*esct) { case '*': if (arg_complete) { deftype = ROFFDEF_USER | ROFFDEF_PRE; @@ -1434,15 +1449,15 @@ roff_res(struct roff *r, struct buf *buf, int ln, int break; } ctx = r->mstack + r->mstackpos; - npos = stesc[2] - '1'; + npos = esct[1] - '1'; if (npos >= 0 && npos <= 8) { res = npos < ctx->argc ? ctx->argv[npos] : ""; break; } - if (stesc[2] == '*') + if (esct[1] == '*') quote_args = 0; - else if (stesc[2] == '@') + else if (esct[1] == '@') quote_args = 1; else { mandoc_msg(MANDOCERR_ARG_NONUM, ln, @@ -1504,7 +1519,7 @@ roff_res(struct roff *r, struct buf *buf, int ln, int } if (res == NULL) { - if (stesc[1] == '*') + if (*esct == '*') mandoc_msg(MANDOCERR_STR_UNDEF, ln, (int)(stesc - buf->buf), "%.*s", (int)naml, stnam); @@ -1532,6 +1547,103 @@ roff_res(struct roff *r, struct buf *buf, int ln, int } /* + * Parse a quoted or unquoted roff-style request or macro argument. + * Return a pointer to the parsed argument, which is either the original + * pointer or advanced by one byte in case the argument is quoted. + * NUL-terminate the argument in place. + * Collapse pairs of quotes inside quoted arguments. + * Advance the argument pointer to the next argument, + * or to the NUL byte terminating the argument line. + */ +char * +mandoc_getarg(char **cpp, int ln, int *pos) +{ + char *start, *cp; + int quoted, pairs, white; + + /* Quoting can only start with a new word. */ + start = *cpp; + quoted = 0; + if ('"' == *start) { + quoted = 1; + start++; + } + + pairs = 0; + white = 0; + for (cp = start; '\0' != *cp; cp++) { + + /* + * Move the following text left + * after quoted quotes and after "\\" and "\t". + */ + if (pairs) + cp[-pairs] = cp[0]; + + if ('\\' == cp[0]) { + /* + * In copy mode, translate double to single + * backslashes and backslash-t to literal tabs. + */ + switch (cp[1]) { + case 'a': + case 't': + cp[0] = '\t'; + /* FALLTHROUGH */ + case '\\': + pairs++; + cp++; + break; + case ' ': + /* Skip escaped blanks. */ + if (0 == quoted) + cp++; + break; + default: + break; + } + } else if (0 == quoted) { + if (' ' == cp[0]) { + /* Unescaped blanks end unquoted args. */ + white = 1; + break; + } + } else if ('"' == cp[0]) { + if ('"' == cp[1]) { + /* Quoted quotes collapse. */ + pairs++; + cp++; + } else { + /* Unquoted quotes end quoted args. */ + quoted = 2; + break; + } + } + } + + /* Quoted argument without a closing quote. */ + if (1 == quoted) + mandoc_msg(MANDOCERR_ARG_QUOTE, ln, *pos, NULL); + + /* NUL-terminate this argument and move to the next one. */ + if (pairs) + cp[-pairs] = '\0'; + if ('\0' != *cp) { + *cp++ = '\0'; + while (' ' == *cp) + cp++; + } + *pos += (int)(cp - start) + (quoted ? 1 : 0); + *cpp = cp; + + if ('\0' == *cp && (white || ' ' == cp[-1])) + mandoc_msg(MANDOCERR_SPACE_EOL, ln, *pos, NULL); + + return start; +} + + +/* * Process text streams. */ static int @@ -3148,7 +3260,7 @@ roff_EQ(ROFF_ARGS) assert(r->eqn == NULL); if (r->last_eqn == NULL) - r->last_eqn = eqn_alloc(r->parse); + r->last_eqn = eqn_alloc(); else eqn_reset(r->last_eqn); r->eqn = r->last_eqn; @@ -3182,7 +3294,7 @@ roff_TS(ROFF_ARGS) mandoc_msg(MANDOCERR_BLK_BROKEN, ln, ppos, "TS breaks TS"); tbl_end(r->tbl, 0); } - r->tbl = tbl_alloc(ppos, ln, r->parse, r->last_tbl); + r->tbl = tbl_alloc(ppos, ln, r->last_tbl); if (r->last_tbl == NULL) r->first_tbl = r->tbl; r->last_tbl = r->tbl; @@ -3659,7 +3771,7 @@ roff_userdef(ROFF_ARGS) ctx->argv = mandoc_reallocarray(ctx->argv, ctx->argsz, sizeof(*ctx->argv)); } - arg = mandoc_getarg(r->parse, &src, ln, &pos); + arg = mandoc_getarg(&src, ln, &pos); sz = 1; /* For the terminating NUL. */ for (ap = arg; *ap != '\0'; ap++) sz += *ap == '"' ? 4 : 1;